The first stages of human visual processing proceed very rapidly, representing the retinal input with high temporal resolution. However, it appears that the visual system extracts only a very few aspects of the retinal input before subsequent slow filtering. The proposed experiments are designed to determine exactly what these privileged, high-speed aspects are. In addition, the transition or transitions from fast to slow processing will be functionally localized with respect to a wide variety of visual processes. Finally, we will test the hypothesis that a common filter is responsible for the low resolution of the slow visual processes.